mablopoule
The process of creation is often a back-and forth between that fuzzy concept in your head (or an natural-language spec that is often ambiguous), and the (usually reduced) physical object / drawing / computer program you've created to materialize this idea. That imperfect projection of your initial idea then helps you refine what this idea implies, and what are the different tradeoff you could take [0].
That back-and-forth has to be made when writing the computer code, even if it's going to be compiled after, because code needs to be detailed enough to force you into expressing the desired behavior with clarity. Prompting Claude on the other hand usually means that you let the LLM resolve this ambiguity for you.
In some case the details don't matter, and it's a productivity gain, and in other case the devils are in the details, and you've let Claude make important technical decision for you, even though Claude doesn't know how much those decision match the needs of the project.
Personally I think of my role as a software developer as someone who know technical possibility (and technical limitations), and weight the different tradeoff it implies to propose code that solves a business problem. And that's the crux of the issue with only creating code by prompting Claude Code: it gradually robs me of my ability to understand the technical possibility and limitation of the system.
[0] This idea has largely been taken from the very good book "Sketching User Experiences: Getting the design right and the right design", about the processus of ideation / sketching.
For me this is the big one. When you code by hand something (especially at the start of a project / feature), it gives you times and opportunity to evaluate different approach, like data-structure or overall architecture. When you're given agentic-generated code to review, those choices are already been made implicitly by Claude, and while it give you the impression that you understand the codebase, you actually haven't been able to process the tradeoff.
Of course, not 'hand-generated' code has the same ROI, and when you're about to do something derivative of something you already understand, agentic development is truly a huge productivity gain.
For me this is the difference between 'vibe coding' and 'coding with AI': Am I able to predict in advance the code that Claude is going to generate, and more importantly, am I able to be surprised when Claude doesn't generate the code I thought it would have (sometime for good technical reasons), or am I just looking at plausible-looking code diffs, and accepting thing without being able to understand the impact on the rest of the codebase. This for me is way more important than what percentage of code lines have been generated by AI.
I've had the opportunity to try both approach: one where I was given a vibecoded project and told to maintain it with Claude: I knew the details of some features, but haven't worked with the codebase enough to really understand the architecture, or be able to mentally reason about the codebase when I'm asked a question during a meeting.
On the another re-implementation of the same product, I've been given more leeway to develop how I wanted, so I've handcrafted the first few features in a transversal way (handling the file architecture, the routing, the way data and components are splitted, and how the API calls are handled), and then I've used Claude to generate the other feature (most of them I knew to be derivative of the first one). In the end something like 90% of the code is generated by Claude, but the end result is both much simpler because I knew where things should be simple and where edge case where, and most importantly I trust what I've put in production.
https://how.complexsystems.fail/
and one of it's associated HN discussion: https://news.ycombinator.com/item?id=32895812
> “The merchants of complexity will try to convince you that you can’t do anything yourself these days,” wrote David Heinemeier Hansson (DHH), the creator of Ruby on Rails. “You can’t do auth, you can’t do scale, you can’t run a database, you can’t connect a computer to the internet. You’re a helpless peon who should just buy their wares. No. Reject.” [1]
DHH also did a very inspiring talk about mastery and why he loved the Ruby language in the "DHH is right about everything" [2] video.
[1] https://thenewstack.io/developers-rail-against-javascript-me...
React shines when it comes to the composability of components, eg. for data-table with customizable rows. It's still possible in Angular, but it's heavier.
For the rest (syntax, the ecosystem, routing, data-handling, DI), Angular is so much more straightforward. Services with dependency-injection is 99% of the time way simpler to reason about than React hooks, especially when you start to need cascading changes between hooks after user interaction.
I still use RxJS, but mostly in the top-level component and/or service who orchestrate between data, url state and api responses. Those top-level page usually keep the default change detection instead of the 'on-push' strategy).
Of course don't fraud by like pretending you're a statistician when you have absolutely no mathematical background, but also don't take at face value the "Must have {x} years of experience in {y} tech" requirement when you know you have the necessary work experience to have a good grasp on it in a few weekend prototypes, and you also know that the job doesn't actually require deep expertise of that particular tech.
I did the same for my first React.js job, and I didn't feel bad because 1) I was honest about it and did not sold myself as a React expert, and 2) I had 10 years of front-end development, and I understood web dev enough to not be baffled by hooks and the difference between shallow copy vs. deep copy of a data structure, so passing technical test was good enough for it.
The linked page in the thread is short and quite enlightening, but here is the relevant passage:
> Rule 9: Human operators have dual roles: as producers & as defenders against failure.
> The system practitioners operate the system in order to produce its desired product and also work to forestall accidents. This dynamic quality of system operation, the balancing of demands for production against the possibility of incipient failure is unavoidable. Outsiders rarely acknowledge the duality of this role. In non-accident filled times, the production role is emphasized. After accidents, the defense against failure role is emphasized. At either time, the outsider’s view misapprehends the operator’s constant, simultaneous engagement with both roles.
There's plenty of guru who say that they are the reincarnation of Jesus and/or Buddha, doesn't mean that we have to take their word for it.
In the same vein, North Korea is officially the "Democratic People's Republic of Korea", even though it's obviously not a democracy.
Framework can be more or less modular, Angular or Ember choose to be 'battery included', while React choose to be more modular, which is simply choosing the other end of the spectrum on the convenience-versus-flexibility tradeoff.
React ostensibly only care about rendering, but in a way that force you to structure your whole data flow and routing according to its rules (lifecycle events or the 'rules of hooks', avoiding mutating data structures); No matter what they say on the official website, that's 100% framework territory.
Lodash or Moment.js, those are actual bona fide libraries, and nobody ever asked whether to use Vue, Angular or Moment.js, or what version of moment-js-router they should use.
Before Johannes Kepler had the insight of describing the orbits of the planets with ellipsis, peoples were using the (conceptually simpler) circles which didn't completely match the observed movement of celestial body such as Mars, thus resulted in complicated circle-within-circles orbits to try to model reality. By introducing a more complex basic shape (ellipsis instead of circle) which happened to match the underlying reality more, the overall description of orbits got greatly simplified.
It's a phenomenon I've seen a few time in my career so far: that while often there's complex code because there are actually complex hedge case to handle (essential complexity), sometime it's really because the data structure used to model the thing you're handling is slightly missing the mark, making things fit almost-but-not-quite, and many operation done around to handle data can be greatly simplified (if not avoided altogether) by changing the underlying data-structure.
(Also, Alan Kay apparently did another talk called "Is it really complex, or did we just make it complicated"[2] that seems pertinent to the thread, though I haven't watch it yet)
[1] https://www.youtube.com/watch?v=NdSD07U5uBs [2] https://www.youtube.com/watch?v=ubaX1Smg6pY
Funny thing is, at first it was the other way around! 'Computer Vision' has always been a sub-field of AI, but the term was more widely used by academics during a previous AI winter as a way to avoid the tainted 'AI' label.
It's not about drinking the corporate kool-aid, but about taking pride in what you've put in the world (even potentially as a hobby), having a sense of craftsmanship, or even maintaining a certain work ethic.
Even the "making money" part can be tied to a very deep sense of providing for your loved ones, and a sense of personal responsibility.
As a french-speaking person, I now find myself seeing french youtubers seemingly posting videos with english titles and robotic voice, before realizing that it's Youtube being stupid again.
What's more infuriating is that it's legitimately at heart a cool feature, just executed in the most brain-dead way possible, by making it opt-out and without the ability to specify known languages.
It's very well written, and it's the first time I've physically reacted that much to the awkwardness of fictional characters.
On a more serious note, this is also why I'm wary of the "punching up" or "punching down" rethoric, because it's often easy to downplay any form of violence as justified retribution.
To quote Joel Spolsky, "When you’re working on a really, really good team with great programmers, everybody else’s code, frankly, is bug-infested garbage, and nobody else knows how to ship on time.", and that's the state we end up if we believe in the hype and use LLMs willy-nilly.
That's why people are annoyed, not because LLMs cannot code like a senior engineer, but because lots of content marketing a company valuation is dependent on making people believe it's the case.
My theory is that there was a perfect storm around 2015 where everyone and their dog was encouraged to learn to code, especially by going through a coding bootcamp where they were mainly taught Javascript and React. At the same time there was a general enthusiasm for Open-Source, and of using Github as a sort of alternative, better Linkedin in order to get your first job as a software engineer.
As a result lots of silly packages were created (and used !) by well-meaning junior developers, who were told that coding is very simple but also fraught with peril, so if they are serious then they better should use packages such as 'is-odd' which is clearly more professional than doing it yourself, cause it follows the DRY principle and also get updated test by lots of people, etc...
Something like:
<header class="main-header">
<input class="hamburger" id="hamburger-action" type="checkbox" />
<label for="hamburger-action">Some icon here</label>
<nav class="main-navigation">My menu here</nav>
</header>
The link is about a grid system with different mathematical properties which is used at Uber.